caffeSigmoidCrossEntropyLossLayer

代码&＃xff1a;

layer {name: "loss_2"type: "SigmoidCrossEntropyLoss"bottom: "score_2"bottom: "label"top: "loss_2"loss_weight: 0.1
}

SigmoidCrossEntropyLossLayer的输入bottom[0]&＃xff0c;bottom[1]&＃xff0c;其中bottom[0]是输入的预测的结果&＃xff0c;bottom[1]是标签值。bottom的维度都是(N×C×H×W)(N×C×H×W)&＃xff0c;bottom的表示符号是x&＃xff0c;bottom[1]是p∈[0,1]